Editor’s Note: This is the nineteenth in a series of posts on Book of Mormon stylometry. For other entries in this series, see the links below.
The earliest study of Book of Mormon stylometry was published in 1981 by Wayne A Larsen, Alvin C Rencher and Tim Layton. What we are going to see as we look into it is that we, as relatively casual readers of stylometry, know more than they did about possible pitfalls than they did when they were actively researching stylometric problems. But despite this and the numerous criticisms that were directed at their work, certain prominent conclusions from Larsen and coworkers have held up very well to the test of time.
“Who Wrote the Book of Mormon? An Analysis of Wordprints” Larsen, Rencher, and Layton, BYU Studies, 20:3 (not open access)
In this earliest work, Larsen et al. divided up the Book of Mormon as much as possible by author based on internal evidences of the text. Much of this was simple, but sometimes great attention was required to separate rapid changes in voice. For example, if Mormon was quoting someone else in the middle of a passage, those words needed to be removed from Mormon’s sample. Again, note that they took the text at face value for deciding how to assign authorship. Of the over 100 people who spoke or wrote in the Book of Mormon, 22 authors contributed 1000 words or more. These authors were used for statistical comparisons. Larsen et al. also included the following authors and sources as controls and for additional testing: Sidney Rigdon and Solomon Spalding, other known works by Joseph Smith, and works by W. W. Phelps, Oliver Cowdery, and Parley P. Pratt, the Lectures on Faith, two sections from the Doctrine and Covenants, and an anonymous article from Times and Seasons. They were careful to use only writings directly attributable to Joseph Smith, and not recorded through scribes.
Larsen et al. used three different measures of stylometric features still in use today: letter frequencies, common non-contextual word frequencies, and rare non-contextual word frequencies. They used sophisticated yet established statistical methods to make the comparisons. So far, so good. Everything seems like normal stylometry. Let’s look at their results.
At this point I’ll say, if you are a believer that the Book of Mormon was written by ancient prophets and faithfully (at least mostly) translated by Joseph Smith, go read the paper. Larsen et al. observe some subtle and interesting findings that probably aren’t statistically significant at the 95% level, or else probably lack sufficient controls to truly be proven. Things like 11 distinct authors, contemporaries having more similar styles than authors who lived hundreds of years apart, and other similar conclusions. But Larsen et al. were working under the false assumption that the passage of time, changes in subject matter, and changes in form did not affect stylometric measures. It is true that their effect is often unobservable, but enough cases have been uncovered since 1981 that researchers no longer feel they can take these things for granted. Despite these caveats, there are some pronounced and interesting observations that have been upheld, at least qualitatively, by later studies.
From this first figure we see the now familiar conclusion that the works written in the 19th century by known authors had different stylometric features from the Book of Mormon. In other words, we don’t know who the Book of Mormon authors were. Now how about within the Book of Mormon?
Within the Book of Mormon, we find authors with distinct, but overlapping, styles—a lot like the Schaalje and Fields studies showed for the 19th century candidate authors. Since I love editorializing, I’m still a bit amazed that Holmes didn’t admit to this same phenomenon in his Book of Mormon paper. It took me no work at all to draw circles analogous to these around the different Book of Mormon authors on his plot, and the distributions of features matched much better all of the control authors than Holmes’s own, monster, “prophetic voice” grouping. I keep thinking back to Morton’s statement about things being decided for statisticians, but the general public resisting even 99.9999% certainty if it doesn’t match their biases. Except with the Book of Mormon, it’s even some statisticians resisting their own evidence and espousing statistically crazy theories to explain the evidence—like Joseph Smith creating a “prophetic voice” with wider stylometric variation than has ever been reported for a single author—even Faulkner. Or “dual authorship” signals with no dual authorship controls and significant evidence for at least four authors.
The Book of Mormon was written by many authors. We don’t know who the authors were. Whatever your theory for how the Book of Mormon came to be had better deal with these facts or it just doesn’t match reality. That doesn’t mean every theory that can accommodate these facts is necessarily true, but this is what you have to deal with if you claim to care about evidence.
I’ll try out some hypotheses myself, but first we need to confirm that the Book of Mormon is as stylometrically unique as I’ve claimed, and also look at clear evidence that Book of Mormon language contains much from the early 19th century.